MFS is an autosomal-dominant systemic disorder of connective tissue with an estimated prevalence of 1 in 5,000–10,000 individuals⁶. The condition is caused by mutations in FBN1, the gene encoding the extracellular matrix protein fibrillin-1 (ref. 6). Clinical manifestations of MFS include bone overgrowth, ocular lens dislocation, emphysema and cardiac complications such as aortic aneurysm. The majority of individuals with MFS lack the ability to increase muscle mass in response to physiologic signals for hypertrophy including growth and exercise. A subset of individuals, prominently those with neonatal onset of severe and rapidly progressive MFS, has profound muscle hypoplasia and hypotonia throughout life. Fibrillin-1 is a structural component of the extracellular matrix microfibril that has been shown to negatively regulate TGF-β activation and signaling³. Analyses of fibrillin-1–deficient mice showed that enhanced activation of and signaling by TGF-β contributes directly to failed distal alveolar septation, myxomatous changes of the atrioventricular valves and aortic aneurysm formation^2,3,7.

TGF-β belongs to a family of cytokines that transduce their signal through the SMAD intracellular signaling cascade⁸. In skeletal muscle, in vitro evidence suggests that TGF-β impairs myocyte differentiation during myogenesis^1,5,9. In addition, TGF-β has been thought to be involved in the formation of fibrosis in response to injury, inflammation or disease^9–11. But a pathogenetic role for TGF-β in other mechanistic aspects of inherited myopathic disorders has not been shown. To investigate a potential role for TGF-β signaling in the development of myopathy in fibrillin-1–deficient mice, we analyzed animals carrying a targeted mutation (C1039G) in exon 25 of the mouse Fbn1 gene¹². This mutation is representative of the most common class of mutations causing human MFS; that is, substitutions of cysteine residues in the EGF-like domains of fibrillin-1. Mice homozygous for the C1039G mutation (Fbn1^{C1039G/C1039G}) have an emaciated appearance and die between 10 and 14 d of age secondary to aortic dissection¹². Analysis of 10-d-old wild-type and homozygous mutant litter mates showed a significant (P < 0.005) discrepancy in body weight that correlated with architectural abnormalities in all skeletal muscle groups examined (Supplementary Fig. 1 online), including a significant decrease in muscle fiber size (18.5 ± 2 versus 13 ± 1.5 μm; P < 0.001) and number (1,504 ± 26 versus 1,380 ± 25 fibers per muscle; P < 0.001) and increased amounts of interstitial tissue and fat between muscle fiber bundles in mutant mice. Mice heterozygous for the C1039G mutation (Fbn1^C1039G/+) recapitulate the dominant nature of human MFS and show phenotypic manifestations in the pulmonary, cardiovascular and skeletal systems¹². Immunohistochemical staining for fibrillin-1 showed decreased endomysial expression in skeletal muscle from Fbn1^C1039G/+ mice and humans with MFS (Supplementary Fig. 1). We observed marked variation in fiber size and endomysial fibrosis in both sets of samples (Fig. 1a and Supplementary Fig. 1).

Open in a separate window

Figure 1

Evaluation of steady-state and regenerating skeletal muscles in fibrillin-1 deficient mice. (a) Hematoxylin and eosin staining of quadriceps muscle (upper panels) shows marked variation of fiber size in Fbn1^C1039G/+ mice. Note several small and split fibers (asterisks), fibers with central nucleation and endomysial thickening. TGF-β antagonism in vivo reverses myopathic architecture in Fbn1^C1039G/+ mice. Increased TGF-β signaling, as evidenced by nuclear accumulation of pSmad2/3 (middle panels) and periostin expression (lower panels) in Fbn1^C1039G/+ mice, when compared to wild-type mice or Fbn1^C1039G/+ mice treated with TGF-β–neutralizing antibody. Analysis of the cross-sectional area (myofiber CSA in μm²) of tibialis anterior muscle fibers shows a decrease in fiber size in Fbn1^C1039G/+ mice when compared to wild-type mice or Fbn1^C1039G/+ mice treated with TGF-β–neutralizing antibody (graph). Scale bars, 70 μm (low magnification) and 50 μm (high magnification, inset boxes). (b) Impaired muscle regeneration in Fbn1^C1039G/+ mice. Few newly-formed muscle fibers and disorganized muscle architecture with numerous small fibers (arrows) 4 and 18 days after cardiotoxin-induced muscle injury, respectively, in Fbn1^C1039G/+ mice, when compared to wild-type mice or Fbn1^C1039G/+ mice treated with TGF-β–neutralizing antibody (top panels). Increased TGF-β signaling, as evidenced by increased nuclear accumulation of pSmad2/3 and periostin expression in Fbn1^C1039G/+ mice, when compared to wild-type mice or mice treated with TGF-β–neutralizing antibody (bottom panels). Morphometric analyses of tibialis anterior muscle 18 d after cardiotoxin injection shows reduced myofiber CSA (in μm²) in Fbn1^C1039G/+ mice, when compared to wild-type mice or Fbn1^C1039G/+ mice treated with TGF-β–neutralizing antibody (graph). Scale bars, 40 μm.

Ligand-activated TGF-β receptors induce phosphorylation of Smad2 and Smad3 (hereafter referred to as pSmad2/3), which form heteromeric complexes with Smad4 that translocate to the nucleus and mediate target gene responses⁸. We observed nuclear accumulation of pSmad2/3 in myofibers of Fbn1^C1039G/+ mice as compared to wild-type littermates (Fig. 1a and Supplementary Fig. 1), but no nuclear staining in phosphatase-treated tissue sections (data not shown). Further evidence for increased TGF-β signaling derived from analysis of the expression of periostin, a protein known to be induced by TGF-β and expressed in regenerating skeletal muscle¹³. In contrast to wild-type mice, Fbn1^C1039G/+ mice showed high expression of periostin in the sarcolemma and endomysial connective tissue of mature and uninjured skeletal muscle (Fig. 1a).

To show a cause-and-effect relationship between excess TGF-β signaling and development of myopathy in Fbn1^C1039G/+ mice, we achieved systemic TGF-β antagonism in vivo by intraperitoneal injection of 1 mg/kg or 10 mg/kg neutralizing antibody to TGF-β beginning at 7 weeks of age. Histologic and morphometric assessment showed a decrease in fiber size in Fbn1^C1039G/+ mice (mean fiber size, 1,698 ± 49 μm²) when compared to wild-type mice (2,622 ± 55 μm²) that is restored upon treatment with neutralizing antibody to TGF-β (2,443 ± 41 μm²; P < 0.005 and P < 0.008 for Fbn1^C1039G/+ mice versus wild-type and Fbn1^C1039G/+ mice versus TGF-β–neutralizing antibody–treated mice, respectively). Moreover, we observed neither nuclear accumulation of pSmad2/3 nor increased periostin staining in Fbn1^C1039G/+ mice treated with neutralizing antibody (Fig. 1a). These data indicate a causal relationship between dysregulated TGF-β signaling and development of skeletal muscle myopathy in a mouse model of MFS.

The presence of multiple atrophic and split fibers in skeletal muscle of fibrillin-1–deficient mice and individuals with MFS is suggestive of abnormal and/or incomplete muscle regeneration (Fig. 1a and Supplementary Fig. 1). To further investigate the potential mechanism by which TGF-β causes the development of skeletal myopathy, we evaluated the response of skeletal muscle to injury induced by snake venom cardiotoxin. Analysis of wild-type skeletal muscle 4 d after cardiotoxin injection showed numerous regenerating muscle fibers (Fig. 1b). In contrast, skeletal muscle of Fbn1^C1039G/+ mice showed delayed regeneration and only scattered newly formed muscle fibers (Fig. 1b).

At 18 d after cardiotoxin injection, wild-type mice showed substantial muscle regeneration, whereas Fbn1^C1039G/+ mice had multiple small fibers with focal areas of fibrosis, indicative of abnormal muscle repair (Fig. 1b). Systemic administration of neutralizing antibody to TGF-β at the time of and 2 weeks after cardiotoxin injection markedly improved the muscle regeneration capacity of the tibialis anterior muscle at 4 and 18 d after injury in Fbn1^C1039G/+ mice (Fig. 1b). Analysis of the cross-sectional myofiber area of tibialis anterior muscle 18 d after cardiotoxin injection showed a reduction in mean fiber size in Fbn1^C1039G/+ mice (1,145 ± 69 μm² versus 2,389 ± 51 μm² in wild-type mice) that was rescued upon treatment with neutralizing antibody to TGF-β (2,092 ± 47 μm²; P < 0.005 for Fbn1^C1039G/+ mice versus wild-type, and P < 0.006 for Fbn1^C1039G/+ mice versus TGF-β–neutralizing antibody–treated Fbn1^C1039G/+ mice; Fig. 1b). Wild-type and both treated and untreated Fbn1^C1039G/+ mice showed nuclear accumulation of pSmad2/3 and high periostin expression 4 d after cardiotoxin injection (data not shown). These findings are consistent with the prior observation of a transient increase in expression of TGF-β1 and periostin within the first 5 d after muscle injury¹³. In contrast, nuclear accumulation of pSmad2/3 and periostin expression persisted in Fbn1^C1039G/+ mice 18 d after injury; this was not observed in wild-type mice and Fbn1^C1039G/+ mice treated with antibody to TGF-β (Fig. 1b). These data suggest that exaggeration or prolongation, or both, of the physiologic spike in TGF-β signaling that attends muscle injury and repair can limit regeneration and culminate in myopathy.

We next investigated whether increased TGF-β signaling alters the performance of satellite cells, a population of cells responsible for muscle regeneration. Analysis of c-met, a marker for quiescent satellite cells, did not show any significant (see Supplementary Fig. 2 online) difference in the number of satellite cells in wild-type and Fbn1^C1039G/+ mice before injury (Supplementary Fig. 2). In contrast, immunohistochemical assessment of M-cadherin, a marker for proliferating satellite cells¹⁴, at 48 h after toxin-induced injury showed fewer positive satellite cells in the tibialis anterior muscle of Fbn1^C1039G/+ mice (15 ± 5 cells per field) when compared to wild-type mice (42 ± 8 cells per field) or Fbn1^C1039G/+ mice treated with antibody to TGF-β (35 ± 8 cells per field; P < 0.004; Supplementary Fig. 2). Myogenin, a myocyte regulatory factor known to be expressed in proliferating and differentiating satellite cells¹⁵, showed a similar decrease in the tibialis anterior muscle of Fbn1^C1039G/+ mice (42 ± 7 positive cells per field) 72 h after cardiotoxin challenge as compared to wild-type mice (63 ± 7 cells per field) or Fbn1^C1039G/+ mice treated with antibody to TGF-β (59 ± 6 cells per field; P < 0.002; Supplementary Fig. 2). Together, these findings indicate that augmented TGF-β signaling causes impaired muscle repair by inhibiting satellite cell proliferation and differentiation.

We then determined whether losartan, an angiotensin II type 1 receptor (AT1) antagonist which has been shown to lead to a clinically relevant antagonism of TGF-β in other disease states including chronic renal disease and cardiomyopathy^16,17, has an impact on muscle in fibrillin-1–deficient mice. Losartan is used widely to treat hypertension, has an exceptional tolerance profile in all age groups and can prevent aortic aneurysm in a mouse model of MFS⁷. Long-term treatment (6 months) with losartan fully normalized steady-state muscle architecture in Fbn1^C1039G/+ mice (wild-type mean fiber size, 2,741 ± 69 μm²; Fbn1^C1039G/+, 1,746 ± 39 μm²; Fbn1^C1039G/+ plus losartan, 2,527 ± 58 μm²; P < 0.008 for Fbn1^C1039G/+ mice versus wild-type, and P < 0.009 for Fbn1^C1039G/+ mice versus losartan-treated Fbn1^C1039G/+ mice, respectively; Fig. 2). Phenotypic rescue correlated with abrogation of TGF-β signaling in mature skeletal muscle and improved muscle function in vivo (Fig. 2 and Supplementary Fig. 3 online). Moreover, we found that administration of losartan before cardiotoxin-induced injury markedly improved muscle regeneration in Fbn1^C1039G/+ mice (Fig. 2).

Open in a separate window

Figure 2

Losartan antagonizes TGFβ and restores muscle architecture and regeneration in Fbn1^C1039G/+ mice. Placebo-treated Fbn1^C1039G/+ mice show altered quadriceps steady-state muscle architecture with smaller fiber size and nuclear accumulation of pSmad2/3, when compared to wild-type or losartan-treated Fbn1^C1039G/+ mice (top panels). Scale bar, 70 μm. Morphometric analyses showed reduced muscle fiber CSA (in μm²) in Fbn1^C1039G/+ mice, when compared to wild-type or losartan-treated Fbn1^C1039G/+ mice (graph). Few newly-formed muscle fibers with low neonatal myosin expression and disorganized muscle architecture 4 and 18 days after cardiotoxin-induced muscle injury, respectively, in Fbn1^C1039G/+ mice, when compared to wild-type or losartan-treated Fbn1^C1039G/+ mice (bottom panels). Scale bar, 40 μm.

We next questioned whether this mechanism of myopathy was strictly relevant to MFS, or whether we had learned something fundamental about muscle biology that would prove relevant to other muscle disorders. There is descriptive evidence for increased TGF-β activity associated with fibrosis in various genetic and acquired muscle disorders, and it has been shown that administration of decorin (a TGF-β antagonist) can reduce TGF-β activity and collagen content in the diaphragm of dystrophin-deficient mdx mice, an animal model for Duchenne muscular dystrophy^11,18. But a pathogenic contribution of increased TGF-β activity to impaired muscle regeneration in these conditions has not been documented. One key mechanism in the pathogenesis of various degenerative myopathies including some forms of muscular dystrophy is a decline in satellite cell performance and muscle regeneration over time^14,19. TGF-β emerged as an attractive candidate mediator of these effects. In keeping with this hypothesis, we found nuclear accumulation of pSmad2/3 and sarcolemmal expression of periostin in skeletal muscle of dystrophin-deficient mdx mice (Fig. 3a). One complicating factor in interpretation of these data is that myostatin (encoded by Mstn), another member of the TGF-β superfamily that suppresses satellite cell activity²⁰, also signals through the pSmad2/3 cascade²¹. We were able to remove this variable by showing an ongoing increase in nuclear pSmad2/3 and periostin expression in mdx/Mstn^−/− animals (Fig. 3a). Myostatin antagonism has been shown to ameliorate the muscle phenotype in dystrophin-deficient mdx mice^22,23; however, in contrast to TGF-β, various lines of evidence have shown that myostatin expression is decreased in muscular dystrophy, perhaps as a component of an inadequate physiologic attempt at compensation²⁴. Thus, whereas therapeutic strategies aimed at myostatin antagonism may provide some benefit by targeting a parallel pathway, TGF-β antagonism targets a pathway that seems to contribute to the pathogenesis of disease.

Open in a separate window

Figure 3

Increased TGF-β signaling contributes to impaired muscle regeneration in mdx mice. (a) Increased nuclear accumulation of pSmad2/3 and sarcolemmal and extracellular matrix expression of periostin in dystrophin-deficient mdx mice (left panels) and mice deficient in both dystrophin and myostatin (mdx/Mstn^−/−; right panels). Scale bar, 60 μm. (b) Reduced new fiber formation and neonatal myosin expression 4 days after cardiotoxin-induced injury in mdx mice, when compared to wild-type mice or mdx mice treated with TGF-β–neutralizing antibody or losartan (top panels). Extensive fibrosis (as evidenced by vimentin staining) and disorganized muscle architecture 18 days after muscle injury in mdx mice, when compared to wild-type mice or mdx mice treated with TGF-β–neutralizing antibody or losartan (bottom panels). The graph shows the percentage of fibrotic area as compared to the total area of muscle tissue 18 days after cardiotoxin injection. *P < 0.007 for mdx mice treated with TGF-β–neutralizing antibody versus untreated mdx mice; **P < 0.005 for losartan-treated versus untreated mdx mice. Scale bars, 40 μm.

We tested the impact of neutralizing antibody to TGF-β and losartan on the regenerative capacity of 9-month-old mdx mice. At 4 d after muscle injury, wild-type mice showed numerous myofibers expressing neonatal myosin, a marker for active regeneration (689 ± 19 neonatal myosin-positive fibers per tibialis anterior muscle, Fig. 3b). Mdx mice showed significant impairment in this response with fewer neonatal myosin-positive fibers (268 ± 12; Fig. 3b). Treatment with either TGF-β–neutralizing antibody or losartan improved regeneration (556 ± 22 and 513 ± 14 positive fibers, respectively; P < 0.002) when compared to untreated mdx mice (Fig. 3b). After 18 d, wild-type mice showed complete regeneration, whereas mdx mice showed large areas of tissue fibrosis. Again, mdx mice treated with neutralizing antibody or losartan showed improved muscle regeneration and diminished fibrosis, as evidenced by reduced vimentin expression (Fig. 3b).

To show that phenotypic benefit from administration of losartan is derived from inhibition of the TGF-β signaling cascade, we performed immunohistologic analysis of target proteins downstream of AT1. Thrombospondin-1 (TSP-1) acts as a potent mediator of angiotensin II–induced TGF-β activation via AT1 (ref. 25). Wild-type mice did not express significant amounts of TSP-1 in skeletal muscle (Fig. 4a). In contrast, we found strong sarcolemmal expression of TSP-1 in the diaphragm and other muscles of mdx mice, whereas losartan-treated mdx mice showed greatly reduced expression of TSP-1. Moreover, treated mdx mice did not show any significant nuclear accumulation of pSmad2 or increased periostin expression (Fig. 4a), indicative of TGF-β antagonism in vivo. In theory, losartan might rescue the muscle phenotype in mdx mice by restoring the dystrophin-glycoprotein complex through antagonism of AT1-driven ubiquitin-mediated pro-teolysis^26,27. Expression studies excluded this hypothesis (Supplementary Fig. 4 online).

Open in a separate window

Figure 4

Losartan decreases angiotensin II-mediated TGF-β signaling and improves muscle function in mdx mice. (a) Immunofluorescence analysis of target proteins downstream of the AT1. The diaphragm of mdx mice shows increased expression of thrombospondin-1 (TSP-1), a potent activator of TGF-β, and increased nuclear accumulation of pSmad2 and sarcolemmal and matrix expression of peripostin, when compared to wild-type or losartan-treated mdx mice. The inset box shows nuclear accumulation of pSmad2 in a connective tissue-rich region of diseased muscle. Scale bar, 100 μm. (b) Long-term losartan treatment attenuated myopathic disease progression in 9 month-old mdx mice. Representative sections of van Gieson–stained diaphragm from wild-type, mdx and losartan-treated mdx mice. Losartan-treated mice showed significantly (P 0.03) less fibrosis (red) than untreated mdx mice. Scale bar, 150 μm. The graphs show quantification of fibrotic area, expressed as percentage of total muscle area (left), and minimal Feret’s diameter variance coefficient (right). (c) In vitro force-frequency relationship of explanted EDL muscle. Isometric tension (g) versus stimulation frequency (1–150 Hz) was reduced in mdx mice at frequencies equal to or greater than 60 Hz, but was fully restored in losartan-treated mdx mice, as compared to wild-type animals (*wild-type versus mdx mice, **losartan-treated versus untreated mdx mice, P < 0.05; left graph). When force was normalized to muscle cross-sectional area (relative tension), losartan-treated mdx and wild-type mice were indistinguishable, whereas untreated mdx mice showed a significant decrease at frequencies of 100 and 150 Hz, when compared to either wild-type or losartan-treated mdx mice, (*wild-type versus mdx mice, **losartan-treated versus untreated mdx mice, P < 0.05; right graph). Representative low-power whole-muscle montages of EDL muscles from two untreated mdx mice (left) showed an overall decrease in muscle size and fiber content and an increase in fibrosis, as compared to two losartan-treated mdx mice (right). Wild-type mice, n = 4; untreated mdx mice, n = 3; losartan-treated mdx mice, n = 4. Scale bar, 150 μm.

We next determined whether long-term treatment with losartan has a beneficial impact on steady-state muscle architecture and muscle function in mdx mice. Losartan treatment commenced in mdx mice at 6 weeks of age, a time of active muscle necrosis associated with cycles of degeneration and regeneration¹⁹. After 6–9 months of treatment with losartan, analysis of various muscle groups including the diaphragm (the most severely affected skeletal muscle in mdx mice^11,19) and gastrocnemius showed significant attenuation of disease progression in mdx mice (Fig. 4b and Supplementary Fig. 4). Evaluation of tissue fibrosis in the diaphragm of treated mice showed a decrease to 18 ± 4% fibrotic area versus 32 ± 5% in untreated mdx mice (Fig. 4b; P < 0.03). Moreover, assessment of the minimal Feret’s diameter variance coefficient, a parameter that provides a sensitive and reliable representation of fiber size heterogeneity and hence dystrophic changes²⁸, indicated a significant improvement in losartan-treated mdx mice when compared to placebo (Fig. 4b; 304.33 ± 12.85 versus 406.17 ± 21.16, respectively; P < 0.007).

In contrast to therapeutic strategies aimed at curing muscular dystrophy through replacement of dystrophin, a protein that contributes to the stability of muscle fibers, strategies aimed at induction of muscle regeneration (for example, TGF-β or myostatin antagonism) will not address the pace of muscle destruction. By allowing muscle regeneration to keep pace with destruction, however, one would anticipate an increase in muscle size, the number of muscle fibers and hence muscle performance. In keeping with this hypothesis, young children with Duchenne muscular dystrophy maintain muscle function despite high creatine kinase levels in serum (indicative of a high rate of muscle fiber destruction). Functional decline correlates with loss of regenerative capacity and progressive fibrotic changes. As predicted by this scenario, there was no significant decrease in creatine kinase levels in losartan-treated mdx mice, when compared to untreated animals (data not shown).

To assess whether there was a functional benefit of losartan treatment in mdx mice, we performed grip-strength testing in vivo. After 6 months of treatment, mdx mice showed increased fore-limb (data not shown) and hindlimb grip strength when compared to untreated mdx mice (Supplementary Fig. 4). In addition, losartan-treated mice showed significantly (see Supplementary Fig. 3) less muscle fatigue in response to repetitive challenge.

We next analyzed physiological properties of explanted extensor digitorum longus (EDL) muscles derived from 1-year-old wild-type, mdx and losartan-treated mdx mice (9 months of treatment). Treated mdx mice showed an increase in body and EDL wet muscle weight and cross-sectional area (Supplementary Table 1 online). When compared to untreated mdx mice, the losartan-induced increase in EDL muscle mass correlated with a significant increase in the number of fibers per muscle, a significant decrease in the minimal Feret’s diameter variance coefficient and no demonstrable increase in single fiber area (Supplementary Table 1). The explanted EDL from losartan-treated mice generated significantly more absolute force over a wide range of stimulation intensities (60–150 Hz) when compared to untreated mdx mice; indeed, the performance was statistically indistinguishable from wild-type mice (Fig. 4c and Supplementary Table 1). At submaximal stimulation frequencies (60 and 80 Hz), the increase in force generation was proportional to the increase in muscle area, probably indicative of increased fiber number. At maximal stimulation frequency (100 and 150 Hz), however, the force generation indexed to cross-sectional area was significantly greater in losartan-treated mdx mice when compared to placebo (Fig. 4c), plausibly reflecting both an increase in fiber number and a decrease in fibrosis.

Excessive TGF-β signaling has been shown to drive pathology in multiple tissues in MFS, albeit by different mechanisms. In skeletal muscle, we showed that increased TGF-β activity impedes the physiologic response of satellite cells to regenerate muscle in multiple genetically defined forms of myopathy, a process essential for the preservation of muscle architecture and performance. Although losartan does not ‘cure’ the disease by restoring dystrophin expression or eliminating muscle membrane fragility, or both, the observations of an enduring ability of muscle regeneration to keep pace with destruction and improved muscle function in the mdx mouse model support speculation that losartan will improve the quality of life and delay death for individuals with Duchenne muscular dystrophy. The benefits of losartan treatment may extend to other states associated with accelerated muscle injury and/or an inadequate response of satellite cells including other inherited or acquired myopathies, catastrophic muscle injury and aging.